1

Quoting a comment from this question: Why did Windows 95 crash the whole system but newer Windows only crashed programs?

"A lot of effort has gone into automatically enforcing that drivers are more robust. This is one of the reasons that Windows 7 seldom bluescreens."

What technologies were used to as windows progressed. I won't restrict this to any particular version, as I think it would be nice to see the timeline of innovation.

Zak
  • 117
  • 2
  • 2
    Just one: Programming :)) But serious, beside that Windows7 isn't realy on topic here, there is no such thing as a 'technology' to be 'used' that's simply marketing snake oil. – Raffzahn Jul 29 '19 at 22:34
  • 2
    With respect to the question in the title: conventional operating system design as practiced from the 1960s onwards: per-process address spaces, object-level access control, privilege restrictions, built on top of a hardware architecture that supports such things (MMU, normal/privileged execution mode, restrictions on certain instructions). With respect specifically to drivers: testing. – dave Jul 29 '19 at 22:37
  • 20 years back, MSN was run on a bunch of NT4 systems, which would crash with sufficient regularity that they had to engineer an elaborate system to extract crash data and ensure machines rebooted in a timely fashion. You have to start with good design for a reliable OS, but the best designs can't save you from bad implementations. Win2K was the first iteration that wasn't completely awful. – fadden Jul 29 '19 at 22:54
  • @another-dave I don't think it's quite that simple. My understanding is that architectural improvements have been put in place to (e.g.) stop video drivers bluescreening the system. That was a major cause of issues in XP but much improved in Vista. – Philip Kendall Jul 29 '19 at 23:22
  • 1
    Microsoft took a look at Unix. Repeat. – Brian H Jul 30 '19 at 02:30
  • 2
    The Amiga had a similar crash reporter called the "Guru Meditation". In WB1.1 it occurred frequently. In later OS versions it got rarer, and I hardly ever see it in WB3.0. What technology did Commodore use to achieve this? None. People just fixed the bugs in their programs! – Bruce Abbott Jul 30 '19 at 02:40
  • Specifically for drivers, and apart from (obviously) just fixing bugs, the answer is a microkernel-like architecture. Real microkernels are still rare and not used in any of the widespread operating systems, but Windows NT at least has a "somewhat similar" design. The key is that drivers are isolated processes, not running directly inside the kernel but communicating with it through IPC. They also should run unprivileged, so they can't do anything to the kernel by accident, but MS left that part out for performance reasons. Anyways, microkernels are not related to retrocomputing... – Felix Palmen Jul 30 '19 at 07:17
  • 2
    @PhilipKendall Graphics drivers were moved from ring 3 in NT 3.x to ring 0 in NT 4.0, for performance reasons. Then at some later date (I think it was done in Vista), they were moved from ring 0 to ring 3 for stability reasons. What goes around, comes around. – user Jul 30 '19 at 09:21
  • This pretty much answers your question: https://www.zdnet.com/article/why-the-blue-screen-of-death-no-longer-plagues-windows-users/ - consider reposting your question elsewhere if you want further detail though – Caius Jard Jul 31 '19 at 05:03

0 Answers0